15.4.1 Content
The focus here is on the content itself, i.e. the data set, including how the data was generated, and what parts of the data is being presented.
Learning Objectives
After completing this tutorial you should be able to
(in short … be pretty good at calling bullsh*t)
You should have already downloaded the directory for this project here. You can open the Rproj for this module either by double clicking on it which will launch Rstudio or by opening Rstudio and then using File > Open Project or by clicking on the Rproject icon in the top right of your program window and selecting Open Project.
Access the quarto document 15_misinformation.qmd and make sure it is in your project directory. Use that file to work through this chapter - you will hand in your rendered (“knitted”) quarto file as your homework assignment. So, first thing in the YAML header, change the author to your name. You will use this quarto document to record your answers.
For this unit, you will not need to load any R packages but do use this document take notes in the document as we discuss misinformation in general and the specific role that visualization play. Make sure that you go back and clean them up for “public consumption”.
In his book ‘On Bullshit’, Harry G. Frankfurt distinguishes between a liar who knows the truth and deliberately tries to convince somebody of something that is untrue and a bullshitter who either does not know the truth or does not care - their focus is solely on persuading the listener. The focus of this distinction is the blatant disregard for truth and a carelessness in how data and facts are used.
In 2015, the National Review posted what they claimed to be “the only climate change chart you need to see”.
1 Remember, when you ‘argue’ you should always state how/why it fulfills or illustrates a definition, statement, or set of criteria.
Climate skepticism comes in different flavors. Early misinformation campaigns were directed at outright denying that global warming is occurring and/or that increased greenhouse gas emissions are the cause, i.e. that we are observing natural climate variability and there is therefore no need for policy to regulate emissions or to change individual behavior.
Arguably, a new flavor of climate skeptics has emerged in the last years that spend less time arguing about whether or not anthropogenic climate change is happening and focus on arguing that the reaction is overblown. Bjorn Lomborg and Michael Shellenberger are very outspoken representatives for this position and focus a lot of their writing on why “climate alarmists” are not only wrong but potentially harmful.
::: {.callout-tip icon=false appearance=“simple”}
As I indicated before, having something that sounds scientific to say when making assertions to laymen is not the same as being correct
~ Christopher Ashley Ford
We will do a deep dive into techniques specifically used by climate deniers & skeptics, with a focuse on the use of visualizations to round out our climate change module.
One of the goals of this course is to equip you with a framework to be able to categorize and assess new information you encounter, or as Neil deGrasse Tyson puts it “To be scientifically literate is to empower yourself to know when someone else is full of bullshit”. The key term here is to empower yourself, i.e. develop a set of skills so that you can pro-actively and habitually assess information - in short, be bullshit resistant.
2 the mis from information?
Here is a non-exhaustive list of strategies that you should tuck away your tool belt for frequent use.
The used cars salesman principle is simple but powerful, when evaluating a source of information sk yourself four simple questions:
This principle is a close relative of Occam’s Razor, also known as the Law of Parsimony.
This principle encompasses both confirmation bias and the dreaded echo chamber - another way of stating it is difficult to not fall in love with your hypothesis and continue to entertain independent evidence.
With ever larger and more available data sets and increasingly complex analysis it is important to not only consider the output and interpretation of an analysis and the methods used (which are becoming increasingly difficult to understand as a layperson) is only ever as good as the data set used in the first place. Consider the source of the data, sample size, how it was generated, and whether it is a legitimate comparison3.
3 Assessing large-scale data sets and how we analyze them is as you may have noticed a central theme of this course. Consider our discussion of responsible conduct in science and the practices driving the reproducibility crisis.
There are two common logical fallacies that can be used to take a relationship of two variables to infer causation - cum hoc ergo propter hoc4 and post hoc ergo proper hoc5.
4 with this, therefore because of this”, i.e. correlation implies causation
5 after this, therefore because of this”, i.e. because something occurred first it must be causing the later
Taking this to ad absurdum6 we have fun examples involving chocolate and Nobel laureates along with storks and babies. Both illustrate the fact that it is important to be careful when inferring causation, consider whether proxies are meaningful, whether there could be a common cause rather than correlation, and if there is a way to design manipulative experiments to determine causation. With increasingly large data sets it is easy to find a “signal” in a noisy data set and spin it into a story7.
6 How many more fancy Latin phrases do I have up my sleeve? We shall probably never truly know… Is this an exempli gratia of using fancy language to obfuscate information? Possibly? Is using pretentious language and jargon also a strategy for spreading misinformation? Also difficult to tell.
7 We will touch on this topic through the course in connection of determining whether your analysis is descriptive, inferential, or causal/mechanistic.
In one of their exhibits, the Pacific Science Center emphasizes four principles for a lay audience to consider when engaging with “science”:
These are important concepts to keep in mind when watching science unfold in real time and out in the open via social media, pre-prints (before peer-review) and in media outlets. It is important to slow down and give science an chance to breath and mature through discussion and testing.
See list of Media Outlet’s ranked by quality of their science reporting; there are several similar assessments you can use as a point of orientiation as well.
“New study finds …” makes for a great headline, though unfortunately occasionally both deliberately and unintentionally by the time the “new study” is translated into a press release and that press release is written up in the media misinformation or disinformation may have sneaked in. This may include
We’ll go into detail on this one in the next section.
Manipulative visualization either present false data or misrepresent data to tell a specific story, i.e. they are either lying with false data or lying with truthful data.
A good way to assess any visualization is to break it down into content, structure, and presentation8.
8 There is some overlap between these categories, you can also think about how these related to the grammar of graphics in having a data set (content), mapping aesthetics and geometry both of which connect to the structure and presentation of the graph.
The focus here is on the content itself, i.e. the data set, including how the data was generated, and what parts of the data is being presented.
Structure refers to how data points are encoded, i.e. what data points represent, what bin sizes are used, how axes are presented etc.9
9 Think about all the components of the grammar of graphics we have discussed learning how to use ggplot.
There is some overlap between Structure and Presentation, but the focus here is more on the visualization as a whole, i.e. this includes not only “scientific-style” figures but other forms of visualization as well.
The presentation for a graph can easily be used for intentional manipulation and misrepresentation of the data.
As we have seen from examples of how visualizations are used in climate change misinformation campaigns, the choice of a scientific(ish) figure is frequently intentional - it lends gravitas to claims. This does not always mean that the visualization meets the standards for a scientific analysis.
Similarly, we have seen that photographs are powerful and can easily be used to be manipulative. Keep in mind, that just because you agree with the statement being supported by a picture that doesn’t mean that picture is being used intentionally to sway you to a point.
Infographics have experienced a rapid growth in popularity in the last ten years10. They are a peculiar mix of data visualization, entertainment, art, and communication. They also vary in the quality of the underlying data set, frequently lack context but feature catchy fun facts and large print key “results”. When done well, they effectively communicate information - unfortunately they are also frequently used for misinformation.
10 Here’s an infographic the rise of infographics for you to refer to.
Two types of potentially misleading figures are Ducks and Glass slippers. Frequently, they fall more into the category of Frankfurtian BS as typically the focus is on how the data is presented using specific gimmicks at the cost of carelessness towards accurately representing the data at
Ducks are named after Big_Duck a building shaped like a duck built to sell duck and duck eggs.
11 If you are familiar with the original version of Cinderella, the Prince and his footman take the glass slipper to all the houses in the kingdom to find his princess. The evil stepsisters chop off their toes and heel in desperation to make it fit. Why this is not in the Disney version remains a mystery. Nevertheless, Glass slipper describes desperately trying to shoehorn data into a visualization type that doesn’t quite fit.
12 For this one “no clue what they are trying to convey” is an option as an answer, but if you are able to figure it out, please do share. Definitely, answer the other parts of the question.
A large component of the general content of this set of activities is inspired by Carl Bergstrom and Jevin West’s course and book on Calling Bullshit: The Art of Skepticism in a Data-Driven World and directly draws on some of their content. Their course is online, including videos, it’s entertaining and informative. Put the book on your Christmas wish list.